11个S2S模式对MJO预报效果的评估分析

利用次季节—季节预报研究计划(Subseasonal to Seasonal Prediction Project, S2S)的多模式产品集,系统评估了产品集中11个模式对MJO的实际预报技巧。如果以距平相关系数ACC为0.5作为有效预报技巧的阈值,S2S各模式的MJO实际预报时效为8～32 d。S2S各模式预报普遍低估了MJO的振幅强度,且预报的MJO传播速度偏慢。通过分析发现,在一个集合预报系统中,集合离散度与均方根误差越接近,它的MJO预报技巧越高。此外,分析S2S各模式MJO预报技巧对起报时间、季节和起报时MJO信号强弱的敏感性发现,当起报时间为冬季且起报时MJO为强信号时,MJO的实际预报技巧较高。

Predictability of the Madden-Julian Oscillation in the subseasonal-to-seasonal prediction models

The prediction skill of MJO was systematically assessed by using ensemble database of the Subseasonal-to-Seasonal (S2S) prediction project. It is found that if the anomaly correlation coefficient ACC is 0.5 that as the threshold value of effective forecasting skills, the actual forecasting time of MJO of S2S models is 8 to 32 days. The prediction skill is also sensitive to its initial conditions. The amplitude intensity of MJO is generally underestimated by S2S models, and the propagation speed of MJO is slow. It is also found that the closer the set dispersion and the root mean square error is, the higher the MJO prediction technique is in a set prediction system. In addition, the sensitivity of MJO forecasting skills of S2S models to the starting time, season and strength of MJO signal was analyzed, which shows that when the starting time is in winter and MJO signal is strong, the actual forecasting skills of MJO are higher.